Assertions and Questionsin Commitment Space Semantics

Manfred Krifka

TaLK ConferenceKeio University, Tokyo

August 7-9, 2016

: 1 / 39

1 Dynamic Semantics and Speech ActsSzabolcsi, Anna. 1982. “Model theoretic semantics of performatives.” Against performative utterances as propositions (Generative Semantics) Against separating semantics (truth-conditionals) from pragmatics (language games) Difference between constatives (descriptives) and performatives:

– describing some state of affairs leaves the state of affairs untouched– performatives change the state of affairs: dynamic transition from one to another.

Example: i + I congratulate you = i′, where i′ like i except that in i′, sp(i) has congratulated addr(i)

Performatives (and speech acts in general) have a semantic type: ⟨s, s⟩ Speech Acts a part of semantics, allowing semantic operators to scope over them.Worked-out proposal: Krifka, Manfred. 2014. “Embedding illocutionary acts.” Transition operator: i′⊶i [φ[i]] ⇔def i′ precedes i immediately ∧ ¬φ[i′] ∧ φ[i]

for all ψ logically independent from φ: ψ[i′] ↔ ψ[i] Here: A particular implementation of this proposal, cf. Krifka 2015,

concentrating on assertions and questions.

Dynamic Semantics and Speech Acts: Dynamic Semantics and Speech Acts 2 / 39

2 A Framework for Illocutionary Acts2.1 Commitment States (CSt)

/ 51 / 69 587 Illocutionary acts change commitments of interlocutors Commitments are represented as propositions Commitments accrue during conversation in Commitment States (CSt)

modeled as sets of propositionsUpdate of commitment state c with speech act Aφ :(1) c + Aφ = c ⋃ {φ},

where φ: the commitment introduced by speech act Aφ.Requirements for update of commitment states: The proposition φ should not be entailed by c

(redundancy; but: increase of saliency, not modeled here) The proposition φ should be consistent with c,

no blatant inconsistencies with salient propositions in c.

A Framework for Illocutionary Acts: Commitment States (CSt) 3 / 39

Figure 1: Update ofcommitment state

c

+ φ

Aφ

2.2 Commitment Spaces (CSp)Common Ground development (Krifka 2008): CG content CG management: Intended continuations of CGNotion of Commitment Space (CSp):(2) C is a CSp iff C is a set of commitment states,

with ⋂C ≠ ∅ and ⋂C ∈ C We call ⋂C the root of C, and write √C. √C is the set of propositions that participants

have positively committed to.Update of C with speech act Aφ:(3) C + A = {c∈C | √C + Aφ ⊆ c}We also use this level to mark the actor or performer of a speech act:(4) C + AS = C +S A = ⟨C + A, S⟩ = [C + A]S,

where S: the person that performs the speech act.If actor is of no concern: Use of wild card.(5) C +* A = [C + A]*

A Framework for Illocutionary Acts: Commitment Spaces (CSp) 4 / 39

Figure 2: Update of CSp: C +S₁ Aφ +S₂ Aψ

√C + ¬φ+ ψ+ φ

+φ,ψ

C

C+Aφ

+ ¬ψ

+φ,¬ψ +¬φ,ψ +¬φ,¬ψ

C+Aφ+Aψ

Aφ

Aψ

S₁:

S₂:

2.3 Boolean Operations: DenegationWhy Commitment Spaces? Boolean Operations: Negation Conjunction DisjunctionModeling of denegation by complementation:(6) I don’t promise to come. (≠ I promise not to come.)Update of a commitment space with denegation of A:

(7) C +S ~A = [C — [C + A]]S

Notice: Denegation is dynamic negation on Commitment Spaces Denegation does not change the root,

no effect on the propositions that the interlocutors are committed to. But denegation has an effect about the possible future development of conversation:

in (monotonic) updates, update with C + A is excluded Cohen & Krifka 2014 call such updates meta speech act.

A Framework for Illocutionary Acts: Boolean Operations: Denegation 5 / 39

Figure 3: Update with Aφ, with A¬φ, and with denegation ~Aφ

√C + ¬φ+ ψ+ φ

+φ,ψ

C

+ ¬ψ

+φ,¬ψ +¬φ,ψ +¬φ,¬ψ

Aφ

C + ~Aφ

C + A¬φ C + Aφ

2.4 Boolean Operations: ConjunctionModeling of conjunction by intersection:(8) C +S [A & B]

= [[C + A] ⋂ [C + B]]S

Always results in a rooted setof commitment states (a Commitment Space)Speech acts generally can be conjoined(cf. Krifka 2001 for quantification and conjunction of questions).Conjunction of Commitment Spaceshas a similar impact as sequential update:(9) C +S [A & B] ≈ [C + A + B]S

See below for sequential update. Anaphoric bindings from first to second conjunct possible with sequential update Sequential update might be a cognitively simpler operation,

and hence preferred over intersection of commitment spaces.

A Framework for Illocutionary Acts: Boolean Operations: Conjunction 6 / 39

Figure 4: Conjunction of regular and meta speech acts

√C +ψ+φ

+φ+ψ

√C A

+ψ+φ

+φ+ψ

B A

B

2.5 Boolean Operations: DisjunctionModeling of disjunction by union.(10) C +S [A V B]

= [[C + A] ⋃ [C + B]]S

Notice: Results in a proper CSp

only for meta speech acts. Speech acts cannot in general be disjoined. Intuitive reason:

It is unclear what the speaker has committed to. Cf. discussion of disjunction of assertions: Gärtner & Michaelis 2010.Fixing non-rooted commitment spaces: If a speech-act operation results in a non-rooted set

of commitment spaces C,add a commitment state c such that C ⋃ {c} is a (rooted) commitment space, such that a proposition p follows from C iff p follows from C⋃c

Here: c = √C ⋃ {φ∨ψ}

A Framework for Illocutionary Acts: Boolean Operations: Disjunction 7 / 39

Figure 5: Disjunction of regular and meta speech acts

√C +ψ+φ

+φ+ψ

√C A

+ψ+φ

+φ+ψ

B A

B

Figure 6: Fixing disjunction

√C

+ψ+φ

+φ+ψ

B A

B

+φ∨ψ

A

2.6 Commitment Space Developments (CSD)Record of the history of the update by a sequence:(11) ⟨C*0, C*1, … C*n⟩, where C*n: the current CSpUpdate of a commitment space development: (12) ⟨..., C*⟩ +S A = ⟨..., C*, [C+A]S⟩ Complete record of the conversation Corresponds to Szabolcsi’s idea of speech acts as world changers And speakers can refer back to order in conversation (As I said at the beginning...)Rejection of last update by rejection operation R (cf. negotiable “table” in Farkas & Bruce 2010):(13) ⟨..., C*, C′*⟩ +S R = ⟨..., C*, C′*, CS⟩, return to next to last CSp, actor: S

Updates as functional applications for CSt, CSp and CSDs:(14) a. Aφ = λc [c⋃φ]

b. AS = λC [{c∈C | √C + A ⊆ c}]S c. AS

= λ⟨..., C*⟩ ⟨..., C*, AS(C)⟩d. RS

= λ⟨..., C*, C′*⟩ ⟨..., C*, C′*, CS⟩

A Framework for Illocutionary Acts: Commitment Space Developments (CSD) 8 / 39

=

3 Assertions3.1 Assertions as commitmentsProposal: By asserting a proposition, speaker makes a public commitment for the truth of that proposition(cf. e.g. Brandom 1983).(15) S ⊢ φ

‘S is publicly committed to / vouches for the truth of φ’Alternative proposal: S wants that addressee believes φ (Bach & Harnish 1979).Problem:(16) Believe it or not, I won the race.But then how does A come to believe φ in typical cases? By committing to a proposition φ, S gives addressee a reason to believe φ. Reason: Committing to false propositions: Social sanctions,which S tries to avoid. As the intention that addressee believes the proposition is cancellable, cf. (16)this is a conversational implicature. General effect of assertion:(17) C* +S₁ S₁⊢φ = [C + S₁⊢φ]S₁

= [{c ⊆ C | √C + S₁⊢φ ⊆ c}]S₁

Assertions: Assertions as commitments 9 / 39

3.2 Syntactic structure of assertionsAssertions involve the following projections: Asserted proposition: TP, Tense Phrase Proposition expressing commitment: CmP, Commitment Phrase Application to CSD (speech act): ActP, Illocutionary Act PhraseFollowing principles of X-bar-syntax(possible rasing of finite verb / subject to CmP, ActP?)(18) [ActP [[Actº . ] [CmP [[Cmº ⊢ ] [TP I won the race]]]]]Compositional interpretation by function ⟦ ⟧S₁S₂, where S₁: Speaker, S₂: Addressee(19) ⟦[ActP [[Actº . ] [CmP [[Cmº ⊢ ] [TP I won the race]]]]]⟧S₁S₂

= ⟦[Actº . ]⟧S₁S₂(⟦[[Cmº ⊢] [TP I won the race]]⟧S₁S₂)= ⟦[Actº . ]⟧S₁S₂(⟦[Cmº ⊢]⟧S₁S₂(⟦[TP I won the race]⟧S₁S₂))with ⟦[TP I won the race]⟧S₁S₂ = ‘S1 won the race’ proposition, TP

⟦[Cmº ⊢]⟧S₁S₂ = λpλS[S⊢p] head of CmP ⟦[Actº . ]⟧S₁S₂ = λRλC*[C + R(S₁)]S₁ head of ActP= λC* [C + S₁⊢‘S₁ won the race’]S₁

A function that updates the last CSp of a CSD, and adds it to the last element.

Assertions: Syntactic structure of assertions 10 / 39

3.3 Reactions to assertionAssertions have two effects: Conventional: Adding speaker’s commitment to proposition Conversational implicature: Adding proposition itself(20) ⟨..., C*⟩ +S₁ S1⊢ φ +S₁ φ

= ⟨..., C*, [C + S1⊢φ]S₁, [C + S1⊢φ + φ]S₁⟩Reactions to assertions:(21) S₁: [ActP [[.] [CmP [[⊢] [TP I won the race]]]]] introduction of propositional

↪ φ discourse referent φ, cf. Krifka 2013S₂: (Okay.) +S₂ φ acknowledgement of φS₂: Yes. +S₂ S₂⊢φ assert φ S₂: No. +S₂ S₂⊢¬φ assert negation of φ, requires retraction

11 / 39

Interlude: Other Speech Act TypesSpeech acts have effects on the world, modeled by proposition that describes it.Veritatives: Public expression of guaranteeing truth; assertions, representatives S adds public commitment to truth of proposition: +S S⊢φ, ‘S vouches for φ’ The proposition φ itself is added by conversational implicatureMutatives: Public expression of change in the world (cf. Barker 2012 on imperatives) Directives, commissives; definitions; declarations; magic spells, prayers; inflectives S restricts the future histories to those in which φ is/becomes true,

e.g. S, to A: Get well! restricts histories to those in which A gets well. Prohibitives as denegations of mutatives, e.g. Don’t get well! exludes those histories Disjunctions as speech act disjunction,

e.g. Eat an apple or eat a pear union of histories in which A eats apple, A eats pear,Get out or I call the police union of histories in which A gets out, S calls the police

In directives, commissives, hortatives: such histories changes result in obligations,perhaps as an indirect speech act, e.g. for directives: +S A !– φ

May also count as expressions of wishes (indirect speech act)Exhibitives: Public display of an attitude or preference: Exclamatives, Optatives (?) S adds a display of an attitude to an entity, a proposition etc. CS: +S S :– φ S, to A: How beautiful this picture is! +S S :– ‘This picture is beautiful’

Interlude: Other Speech Act Types: Interlude: Other Speech Act Types 12 / 39

4 Questions4.1 Questions as meta speech actsQuestions as Common Ground Management: They determine how the common ground should develop Preferred development: Addressee answers the question(22) C* + S1 to S2: Did I win the race?

= [{√C} ⋃ C + S2⊢φ ⋃ C + S2⊢¬φ]S₁

Possible reactions to polar question:(23) a. (22) + S2: Yes. = (22) +S₂ S2⊢φ

b. (22) + S2: No. = (22) +S₂ S2⊢¬φ (24) (22) +S₂ R +S₂ S2: I don’t know. =

⟨..., C*, [{√C} ⋃ C + S2⊢φ ⋃ C ⋃ S2⊢¬φ]S₁, CS₂, [C + S2⊢‘¬S2 knows whether φ’]S₂⟩

13 / 39

4.2 Monopolar questions Polar question as illustrated so far: Offer two assertions, of φ and ¬φ ⇒ bipolar question

The framework also allows for questions that offer just one assertion, of φ⇒ monopolar questions

Candidates for monopolar questions:(25) a. Declarative questions: I won the race?

b. Questions with negated propositions: Did I not win the race?c. Option for regular questions: Did I win the race? (Different from: Did I win the race, or not?)

(26) C* + S1, to S2: I won the race? = [{√C} ⋃ C + S2⊢φ]S₁

Notice that response yes is straightforward, whereas no requires prior rejection Natural way of expressing question bias This option is not available for theories for which questions always denote

a non-singleton set of propositions, or a disjunction, as in Inquisitive Semantics (Roelofson & Farkas 2015).

Questions: Monopolar questions 14 / 39

4.3 Derivation of monopolar questionsMonopolar questions: ActP head ? creates a meta speech act (requests to commit to proposition).(27) ⟦[ActP [[Actº ? Did ] [CmP [[Cmº ⊢ tdid ] [TP I tdid win the race]]]]]⟧S₁S₂

= ⟦[Actº ? ]⟧S₁S₂(⟦[[Cmº ⊢] [TP I did win the race]]⟧S₁S₂)= ⟦[Actº ? ]⟧S₁S₂(⟦[Cmº ⊢]⟧S₁S₂(⟦[TP I did win the race]⟧S₁S₂))with ⟦[TP I won the race]⟧S₁S₂ = ‘S₁ won the race’ proposition ⟦[Cmº ⊢]⟧S₁S₂ = λpλS[S⊢p] head of CmP,

same as assertion ⟦[Actº ? ]⟧S₁S₂ head of ActP, = λRλC*[{√C} ⋃ C + R(S₂)]S₁ applies CmP to addressee= λC*[{√C} ⋃ C + S₂⊢‘S1 won the race’]S₁ monopolar question

Questions: Derivation of monopolar questions 15 / 39

4.4 Commitment Phrases in Conjunct/Disjunct systems (egophoricity)Example: Kathmandu Newari (Hargreaves 2005; cf. Wechsler 2015).(28) Assertions Questions

a. jī: a:pwa twan-ā. d. jī: a:pwa twan-a-la.1.SG.ERG much drink-PST.CJ ‘1.SG.ERG much drink-PST.DJ-Q‘I drank a lot.’ ‘Did I drink a lot?’

b. chā a:pwa twan-a. e. chā a:pwa twan-ā-la. 2. SG.ERG much drink-PST.DJ 2.SG.ERG much drink-PST.CJ-Q

‘You drank a lot’ ‘Did you drink a lot?’c. wā: a:pwa twan-a. f. wā: a:pwa twan-a-la.

3. SG.ERG much drink-PST.DJ ‘3. SG.ERG much drink-PST.DJ-Q ‘he/she drank a lot’ ‘Did he/she drink a lot?’

Proposal: CJ presupposes Committer = Subject, DJ presupposes Committer ≠ Subject(29) ⟦CJ⟧ ̱S₁S₂ = λPλxλS.S=x[S⊢P(x)] ⟦DJ⟧S₁,S₂ = λPλxλS.S≠x[S⊢P(x)]For 3rd pers. subjects in commitment reports; embedded assertions (cf. Krifka 2014):(30) Syām-ā a:pwa twan-ā hã. Syām-ā a:pwa twan-a hã.

Syam-ERG much drink-PST.CJ EVD Syam-ERG much drink-PFV.DJ EVD‘Syam said that he drank too much.’ ‘It is said that Sam drank too much.’

Questions: Commitment Phrases in Conjunct/Disjunct systems (egophoricity) 16 / 39

4.5 Disjunctive questions(31) Did Ed meet Ánn, or did Ed meet Béth? raising accent (question) Proposal: Question disjunction(32) ⟦[ActP [ActP Did Ed meet Ann] or [ActP Did Ed meet Beth]]⟧S₁S₂

with ⟦[ActP Did Ed meet Ann]⟧S₁S₂ = λC*[{√C} ⋃ C + S₂⊢‘Ed met Ann’]S₁

and ⟦[ActP Did Ed meet Beth]⟧S₁S₂= λC*[{√C} ⋃ C + S₂⊢‘Ed met Beth’]S₁

and ⟦or⟧S₁S₂ = λAλA′λC*[A(C) ⋃ A′(C)]S₁, where A, A′: variables over speech acts = λC*[[{√C} ⋃ C + S₂⊢‘Ed met Ann’] ⋃ [{√C} ⋃ C + S₂⊢‘Ed met Beth’]]S₁

= λC*[{√C} ⋃ C + S₂⊢‘Ed met Ann’ ⋃ C + S₂⊢‘Ed met Beth’]]S₁

Questions: Disjunctive questions 17 / 39

Figure 13: Disjunctive question as disjunction of two monopolar questions

. /

√C C

. 0 . /

√C C

. 0 . /

√C C

. 0

S₁:

V =

4.6 Alternative (disjunctive) questionsDisjunctive questions come about as disjunctions of monopolar questions;recall that disjunctions are defined for meta speech acts. (33) S₁ to S₂: Did I win the race, or not?

= ⟦[ActP Did I win the race]⟧S₁S₂ V ⟦[ActP did I not win the race]⟧S₁S₂

= λC*[{√C} ⋃ C + S2⊢‘S1 won the race’] ⋃ [{√C} ⋃ C + S₂⊢¬‘S₁ won the race’]]S₁

Simple answer yes / no avoided, as there are two propositional discourse referents:(34) [ActP [ActP ? Did [CmP ⊢ [IP I win the race]]] or [ActP ? did [CmP ⊢ [IP I not win the race]]]]

↪ φ ↪ ¬φCf. disjunctive formation of bipolar questions in Mandarin:(35) a. monopolar question: b. bipolar question:

Nǐ chī píngguǒ ma? Nǐ chī bù chī píngguǒ? you eat apple QUEST you eat not eat apple ‘Do you eat apples?’, ‘You eat apples?’ ‘Do you eat apples (or not)?’

Questions: Alternative (disjunctive) questions 18 / 39

Figure 14: Disjunction of monopolar questions

..

√C S₁: S₁:

4.7 Constituent Questions as disjunctive questions(36) a. Which woman did Ed meet? (Ann, Beth, or Carla?)

b. Did Ed meet Ann, or did Ed meet Beth, or did Ed meet Carla?In English, wh-phrases in root questions are moved to SpecActP:(37) ⟦[ActP [DP which woman]i [Act′ [Actº ?-did] [CmP [[Cmº ⊢] [TP Ed tdid meet ti]]]]]⟧S₁S₂

= ⟦[DP which woman]⟧S₁S₂ (λxi ⟦[Act′ [Actº ?-did] [CmP [[Cmº ⊢] [TP Ed tdid meet ti]]]]⟧S₁S₂,tᵢ/xᵢ)with λxi ⟦[Act′ [Actº ?-did] [CmP [[Cmº ⊢] [TP Ed tdid meet ti]]]]⟧S₁S₂,tᵢ/xᵢ

= λxi λC*[{√C} ⋃ C + S₂⊢‘Ed met xi’]S₁

and ⟦[DP which woman]⟧S₁S₂ = λR λC*[ U [R(x)(C)]S₁

x∈⟦woman⟧

= λC*[{√C} ⋃ ⋃{C + S₂⊢‘Ed met xi’ | xi∈woman}]S₁

Questions: Constituent Questions as disjunctive questions 19 / 39

Figure 15: Constituent question Which woman did Ed meet? as disjunction of monopolar questions.

. /

√C C

. 0 . 1

S₁:

. /

√C C

. 0 . 1 . /

√C C

. 0 . 1 . /

√C C

. 0 . 1V V =

5 Focus in Answers and Questions5.1 Focus in Answers(38) a. S₁: Who met Ann? S₂: [ED]F met Ann.

b. S₁: Who did Ed meet? S₂: Ed met [ANN]F

Focus in answer leads to a set of alternatives that matches the question (Rooth 1992);here: alternative assertions.(39) ⟦[ActP Ed met [ANN]F.]⟧S₂S₁ (with alternatives Ann, Beth, Carla):

meaning: λC*[C + S₂ ⊢ ‘Ed met Ann’]S₁

alternatives: {λC*[C + S₂⊢‘Ed met Ann’]S₂,λC*[C + S₂⊢‘Ed met Beth’]S₂,λC*[C + S₂⊢‘Ed met Carla’]S₂}

Condition for Q/A focus congruence: Alternatives of Answer ⊆ Meaning of Question

Focus in Answers and Questions: Focus in Answers 20 / 39

Figure 16: (a) Meaning of question, (b) meaning of answer, (c) alternatives of answer

. /

√C C

. 0 . 1

S₁:

. /

√C C

. 0 . 1

S₁:

S₂: S₂: S₂:

. /

√C C

. 0 . 1

S₁:

S₂: S₂: S₂:

5.2 Focus in questionsHere: Focus in monopolar questions.(40) S₁: Did Ed meet [ÁNN]F? S₂: Yes. rising accent

S₂: #No. / No, he met [BETH]F.Focus indicates alternative monopolar question:(41) ⟦[ActP Did Ed meet [ÁNN]F?]⟧S₁S₂ (with alternatives Ann, Beth, Carla)

meaning: λC* [{√C} ⋃ C + S₂⊢‘Ed met Ann’]S₁

alternatives: { λC*[{√C} ⋃ C + S₂⊢‘Ed met Ann’]S₁, λC*[{√C} ⋃ C + S₂⊢‘Ed met Beth’]S₁, λC*[C*, [{√C} ⋃ C + S₂⊢‘Ed met Carla’]S₁}

The union of the question alternatives form the background, which is accommodated; in case question is answered negatively, this background question remains.

Focus in Answers and Questions: Focus in questions 21 / 39

Figure 17: (a) Background, (b) question, (c) rejection, (d) assertion of negated proposition, (e) assertion of other proposition

. /

√C C

. 0 . 1

S₁:

. /

√C C

. 0 . 1

S₁:

. /

√C C

. 0 . 1

S₁:S₂:

. /

√C C

. 0 . 1

S₁:

S₂: . /

S₂:

. /

√C C

. 0 . 1

S₁:

S₂: . /

S₂:

S₂:

6 Questions with Polarity Phrases6.1 Polarity Phrase(42) A:I don’t believe that you won the race.

B: I DID win the race. (Verum focus)Proposed syntactic structure, with Polarity Phrase PolP(43) [PolP I [ [Polº

pol - did] [IP tI tdid win the race]]]]Semantic contribution of pol:(44) a. Meaning: λp[p] (identity function)Redundant, hence always with alternative:

b. Alternatives: {λp[p], λp[¬p]}(45) ⟦[PolP [ [Polº

pol - did] [IP I tdid win the race]]]⟧S₁S₂

= ⟦[Polº pol]⟧S₁S₂(⟦[IP I tdid win the race]⟧S₁S₂)Meaning: ‘S₁ won the race’Alternatives: {‘S₁ won the race’, ¬‘S₁ won the race’}

Q/A congruence to bipolar question:(46) S2: Did you win the race, or not?

S₁: I DID win the race.

Questions with Polarity Phrases: Polarity Phrase 22 / 39

.

√C C

.B

S₂:

.

√C C

.

S₂:S₁: S₁:

.

√C C

.

S₁:S₁:

Figure 18:

(a) Bipolarquestion

(b)

Q/Acongruence:Alternativesof answerfit question

(c)

Answer

6.2 Bipolar interpretations of yes/no questionsWe have analyzed simple yes/no questions as monopolar,but they arguably also have a bipolar reading, e.g. when auxiliary is accented:(47) S1: DID I win the race?We assume: Alternatives of the polarity phrase project to ActP; raising accent on did(48) ⟦[ActP [[Actº

?-DID] [CmPÞ [[Cmº ⊢ tdid] [PolP [[Polº pol- tdid] [ I tdid win the race]]]]]]]]⟧S₁S₂

Meaning: λC*[{√C} ⋃ C + S₂⊢‘I won the race’]S₁

Alternatives: { λC*[{√C} ⋃ C + S₂⊢‘I won the race’]S₁, λC*[{√C} ⋃ C + S₂⊢¬‘I won the race’]S₁}

(49) S₂: Yes, you did. λC*[C + S₂⊢‘S₁ won the race’]S₂

(50) S₂: No, you didn’t. λC*[C + S₂⊢ ¬‘S₁ won the race’]S₂

Requires prior retraction, then assertion of the only alternative left. Question is not quite symmetric, but signals interest in positive and negative answer.

Questions with Polarity Phrases: Bipolar interpretations of yes/no questions 23 / 39Figure 19: (a) Alternatives of question, (b) Question, (c) Rejection, (d) Assertion of remaining alternative

.

√C C

.

S₁:

.

√C C

.

S₁:

.

√C C

.

S₁:

.

√C C

.

S₁:S₂:

7 Negated Questions7.1 Monopolar question with propositional negationNegation part of the proposition, modifier or per NegP:(51) ⟦[ActP [[Actº ? Did ] [CmP [[Cmº ⊢ tdid ] [TP/NegP I [T′/Neg′ not [TP tI tdid win the race]]]]]⟧S₁S₂

= λC*[{√C} ⋃ C + S₂⊢¬‘S1 won the race’]S₁

Notice: This is different from non-negated monopolar question,

bias towards negative answer In standard accounts

(Hamblin, Groenendijk & Stokhof, Roelofsen)non-negated and negated yes/no questions have the same meaning:{p, ¬p} = {¬p, ¬¬p}

Interpretation of responses yes / no is not straightforward, as two propositional discourse referents, φ and ¬φ, are introduced(cf. Krifka 2013, Meijer e.a. 2015).

Negated Questions: Monopolar question with propositional negation 24 / 39

7.2 Monopolar question with high negationHigh negation is interpreted at the level of the commitment phrase, (52) ⟦[ActP[[Actº ? Did] [CmP/NegP [Cm′/Neg′ n’t [CmP [[Cmº⊢] [TP I tdid win the race]]]]]]⟧S₁S₂

= ⟦[Actº ?]⟧S₁S₂(⟦not⟧S₁S₂(⟦⊢⟧S₁S₂ (⟦[TP I did win the race]⟧S₁S₂)))= λC*[{√C} ⋃ C + ¬S2⊢φ]S₁

With this move, S₁ asks S₂ to express non-commitmenttowards the proposition φ.

Notice that adding ¬S₂⊢φ to the CSp precludes commitment to φ, i.e., S₂⊢φ, but is compatible with commitment to ¬φ, i.e., S₂⊢¬φ.

Hence, ¬S₂⊢φ is pragmatically weaker than S₂⊢¬φ: The former proposition does not force S₂ to also committo ¬φ, whereas the latter proposition forces S₂ not tocommit to φ, as it would be incompatible with S₂⊢φ.

Reactions to high negation questions: The TP introduces a discourse referent φ, can be picked up by no, asserts ¬φ. The answer yes requires a rejection of the last move in. The reaction I don’t know does not require a rejection,

as it is compatible with S₂ being not committed to φ.

Negated Questions: Monopolar question with high negation 25 / 39

7.3 Questions of biasA variety of expressing yes/no questions:(53) a. {√C} ⋃ C + [S2⊢φ]S₁

monopolar question{√C} ⋃ C + [S2⊢¬φ]S₁

monopolar question, negated propositionb. {√C} ⋃ C + S2⊢φ ⋃ C + [S₂⊢¬φ]S₁ bipolar questionc. {√C} ⋃ C + [¬S2⊢φ]S₁ high negation question

{√C} ⋃ C + [¬S2⊢¬φ]S₁ high negation question, negated propositionDiscussion of biases: Büring & Gunlogson 2000, Sudo 2013, Gärtner & Gyuris 2016Sudo discusses two different kinds of bias: Evidential bias Epistemic bias

Negated Questions: Questions of bias 26 / 39

Evidential bias:(54) [S₂ enters the windowless computer room, raincoat dripping.]

a. Is it raining?b. # Is it not raining?c. # Is it sunny?d. # Is it raining, or not?e. # Isn’t it raining?f. # IS it raining?

(55) a. Asking the monopolar question S₂⊢φ, if φ is likely, results in a smooth conversation (simple affirmation).

b. Asking the monopolar question S₂⊢¬φ would result in a likely rejection, which should be avoided in smooth communication.c. Would also result in a likely rejection, as sunny → ¬ rainingd. Bipolar questions suggest that φ and ¬φ are equally likely,

if φ is more likely, (a) is to be preferred. e. Checking whether S₂ would refrain from asserting φ is a rather complex move,

appropriate only if φ is controversial. f. Also a bipolar question, focus on auxiliary indicates alternatives λp[p], λp[¬p]

Negated Questions: Questions of bias 27 / 39

Epistemic bias:(56) S₂: You must be starving. You want something to eat?

S₁: Yeah. I remember this place from my last visit. a. Isn’t there a vegetarian restaurant around here? b. (#) Is there a vegetarian restaurant around here?

Explanation of preference of high negation question (a): S₁ checks whether S₂ refrains from committing to the proposition φ,

that is, whether S₂ is willing to add ¬S₂⊢φ to the common ground. Rationale: S₁ has an epistemic tendency favoring φ and is interested

whether the strength of this belief can be increased; S₁ considers S₂ as a possible independent source.

But S₁ does not want to impose the epistemic tendency for φ on S₂ by making asserting ¬φ an easy option, as with the biased question based on S₂⊢φ (b).

(a) does not force S₂ to commit to φ or ¬φ directly, but rather officially invites S₂ to refrain from a commitment for φ. Explains politeness of high negation questions.

(a) makes it easier to answer negatively, by S₂⊢¬φ; strategy of S₁: maximize the chances for S₂ to actually commit to ¬φ. If S₂ against these odds commits to φ, then S₁ can assume that this commitment was not obtained by force.

Negated Questions: Questions of bias 28 / 39

8 Question tagsMatching and reverse question tags (Cattell 1973):(57) You are tired, are you? (58) / I have won the race, haven’t I?

0 I haven’t won the race, have I?

8.1 Matching question tagsSpeech act conjunction of an assertion and a question(59) I have won the race, have I?

C +S₁ [⟦[ActP [ . ] [CmP [⊢] [TP I have won the race]]]⟧S₁S₂ & ⟦[ActP [ ? ] [CmP [⊢] [TP I have won the race]]]⟧S₁S₂]= [[C + S₁⊢φ] ⋂ [{√C}⋃ C + S₂⊢φ]]S₁

The overall effect is that S₁ proposes to S₂ that both S₁ and S₂ are committed to the proposition φ.

That is, S₁ proposes dark central area as new commitment space. S₁ can propose S₂⊢φ because φ is understood as a commitment

that S₂ has already anyway – Cattell: “Voicing a likely opinion by the addressee”. Hence: Evidential bias towards φ

Question tags: Matching question tags 29 / 39

8.2 Reverse question tagsSpeech act disjunction of an assertion and a question(60) I have won the race, haven’t I?

C +S₁ [⟦[ActP [ . ] [CmP [⊢] [TP I have won the race]]]⟧S₁S₂ V ⟦[ActP [ ? have’nt ] [CmP [⊢ ] [TP [[tn’t] [TP I thave won the race]]]]]⟧S₁S₂]

Epistemic bias towards φ, seeking confirmation

Question tags: Reverse question tags 30 / 39

9 Embedded Questions9.1 Nature and kind of embedded questionsQuestions also occur as embedded syntactic objects:(61) a. Who won the race?

b. Bill knows who won the race.But there are important differences between root and embedded questions:(62) a. Who did Ed meet? *Who Ed met?

b. Bill knows who Ed met. *Bill knows who did Ed meet.(63) a. Did Ed meet Beth? *Whether / if Ed met Beth?

b. Bill knows whether / if Ed met Beth. *Bill knows did Ed meet Beth?(64) a. Did Ed meet Ann or Beth? *Whether Ed met Ann or Beth?

b. Bill knows whether Ed met Ann or Beth. *Bill knows did Ed meet Ann or Beth?Discourse particles in German:(65) a. Wen hat Ed denn getroffen?

b. Bill weiß, wen Ed *denn getroffen hat.This is evidence that embedded questions do not involve ActP and CmP, but they involve structure beyond a TP.

Embedded Questions: Nature and kind of embedded questions 31 / 39

9.2 WhetherEmbedded questions and declaratives form a CP, not a CmP or ActP:(66) a. [CP [[Cº whether] [TP Ed met Ann]]]

b. [CP [[Cº that] [TP Ed met Ann]]]Whether / Q turns TP proposition into a set of propositions, with two options: Bipolar: λp {p, ¬p} (cf. etymology: wh + either) Monopolar: λp [p]Evidence for monopolar operator: (67) I doubt whether the Benefits of opposition to the Constitution... (G. Washington)(68) a. [CP [CP [[Cº whether] |TP Ed met Ann]]] or [CP [[Cº whether] [TP he met Beth]]]]

b. [CP [[Cº whether or not] [TP Ed met Ann]]]c. [CP [[Cº whether] [TP Ed met ANN, BETH or CARla]]]

(69) ⟦[[CP [[Cº whether] |TP Ed met Ann]]] or [CP [[Cº whether] [TP he met Beth]]]⟧S₁S₂

= λp{p}[‘Ed met Ann’] ∨ λp{p}[‘Ed met Beth’]= {‘Ed met Ann’} ⋃ {‘Ed met Beth’}, = {‘Ed met Ann’, ‘Ed met Beth’}

Bipolar operator: (70) ⟦[CP [[Cº whether] [TP Ed met Ann]]]⟧S₁S₂

= λp{p, ¬p}(‘Ed met Ann’), = {‘Ed met Ann’, ¬‘Ed met Ann’}

Embedded Questions: Whether 32 / 39

9.3 Embedded Constituent QuestionsAssumption for syntactic structure: Qu head(71) a. [CP whoi [[Cº Qu ] [TP Ed met ti]]]

b. [CP wen [[Cº (dass)] [TP Ed twen getroffen hat]]] (Southern German)Qu is interpreted like whether, i.e. introduces singleton sets.(72) ⟦[CP [which woman]i [[Cº Qu ] [TP Ed met ti]]]⟧S₁S₂

= ⟦which woman⟧S₁S₂(λxi [⟦Qu⟧S₁S₂(⟦[TP Ed met ti]⟧S₁S₂,tᵢ/xᵢ)])with ⟦[TP Ed met ti]⟧S₁S₂,tᵢ/xᵢ = ‘Ed met xi’and ⟦Qu⟧S₁S₂ = λp{p}and ⟦which woman⟧S₁S₂ = λR U R(x) x∈⟦woman⟧S₁S₂

we have: U {‘Ed met x’}, = {‘Ed met x’ | x∈⟦woman⟧S₁S₂} x∈⟦woman⟧S₁S₂

Question-embedding know reduces to proposition-embedding know:(73) ⟦know⟧(Q)(⟦Ed⟧) ⇔ ∀p∈Q[p is true → ⟦know⟧(p)(⟦Ed⟧)]

‘for every true proposition in the set of propositions, Ed knows that it is true.’Notice: strong exhaustive interpretation when Qu is interpreted as λp{p, ¬p}

Embedded Questions: Embedded Constituent Questions 33 / 39

9.4 Comparison: Wh in Root vs. embedded questionsWh in embedded questions: Disjunctions of sets of propositions.(74) a. {p} ∨ {q} = {p} ⋃ {q}, = {p, q}

b. λR [ U R(x)] (λy {p(y)}) = U {p(x)} x∈WH x∈WH

Wh in root questions Disjunctions of functions from CSp to CSp(75) a. λC[A(C)] V λC[B(C)] = λC[A(C) ⋃ B(C)]

b. λR [ U R(x)] (λyλC[A(y)(C)]) = λC [ U A(x)] x∈WH x∈WH

Basic meaning in either case: set union (corresponding to disjunction);difference just a matter of type (where e: entities, st: propositions) Root questions: who is of type [[e → {st}] → {st}] Embedded questions: who is of type [[e → [CSp → CSp]] → [CSp → CSp]]Cf. also: Wh with indefinite interpretation, as in German, or engl. somewhere(76) Ed hat wen getroffen. ‘Ed met someone’(77) a. p ∨ q b. λP U P(x) (λy[p(y)]) who is of type [[e → st] → st]

x∈WH

Embedded Questions: Comparison: Wh in Root vs. embedded questions 34 / 39

9.5 Embedded root questionsPredicates like wonder, ask, be interested in are different: Root syntax possible:(78) a. Ed wondered who he met.

b. % Ed wondered who did he meet. (Irish English, cf. McCloskey 2005) Discourse particles that occur in root questions:(79) a. Wen hat Ed denn getroffen?

b. Ed weiß, wen er *denn getroffen hat.c. Ed fragte sich, wen er denn getroffen hat / habe.

Krifka (2014) argues that such questions are different: They may denote illocutionary acts This is possible, as ActPs are semantic objects, with a proper semantic type

(CSD → CSD)(80) Ed [wondered [ActP who did he meet]](81) x wonders Q, where Q: a question speech act

‘in the situation s referred to, x is interested in the answer to the speech act Q performed in that situation’

Embedded Questions: Embedded root questions 35 / 39

10 ConclusionGoals of the talk: Argue for dynamic semantics for speech acts (cf. Szabolcsi) Introduce a framework of conversation as development of common ground

(cf. Stalnaker, Lewis, ...) Common grounds contain the commitments of interlocutors (Commitment States) New: Common grounds have a projective component (Commitment Spaces)

that models common ground management Questions have an effect on the projective component:

they restrict the legal development of the common ground. There are “monopolar” questions that project just one legal development;

this can be used to model biased questions Proposals for focus in answers to questions and focus in questions,

in particular, focus in polarity questions Proposals for polarity (yes/no) questions, alternative questions,

constituent (wh-) questions, question tags. Explanation of biases of such questions Relation between root and embedded questionsThis talk is partly based on Krifka 2015.

Conclusion: Conclusion 36 / 39

11 ReferencesAsher, Nicholas & Brian Reese. 2007. Intonation and discourse: Biased questions. Interdisciplinary

Studies on Information Structure 8, 1-38. Bach, Kent & Robert M. Harnish. 1979. Linguistic Communication and Speech Acts. Cambridge,

Mass.: MIT Press.Brandom, Robert B. 1983. Asserting. Noûs 17, 637-650.

doi: http://dx.doi.org/10.2307/2215086.Büring, Daniel & Christine Gunlogson. 2000. Aren’t positive and negative polar questions the same?

Talk at Linguistic Society of America (LSA) Annual Meeting.Cattell, Ray. 1973. Negative transportation and tag questions. Language 49, 612-639. doi:

http://dx.doi.org/10.2307/412354.Cohen, Ariel & Manfred Krifka. 2014. Superlative quantifiers and meta-speech acts. Linguistics and

Philosophy 37, 41-90. doi: http://dx.doi.org/10.1007/s10988-014-9144-x.Farkas, Donka F. & Floris Roelofsen. 2015. Polar initiatives and polar particle responses as a window

onto the interpretation of questions and assertions. Language 91, 359-414. doi: http://dx.doi.org/10.1353/lan.2015.0017.

Farkas, Donka F. & Kim B. Bruce. 2010. On reacting to assertions and polar questions. Journal of Semantics 27, 81-118. doi: http://dx.doi.org/10.1093/jos/ffp010.

Gärtner, Hans-Martin & Jens Michaelis. 2010. On Modeling the Distribution of Declarative V2-Clauses: the Case of Disjunction. In: Bab, Sebastian & Klaus Robering, (eds), Judgements and Propositions. Logical, Linguistic, and Cognitive Issues. Berlin: Logos, 11-25.http://semanticsarchive.net/Archive/DY1ZDM1M/paper.pdf

References: References 37 / 39

Gunlogson, Christine. 2002. Declarative questions. SALT XII, 124-134. Ithaca, NY: Cornell University.doi: http://dx.doi.org/10.3765/salt.v12i0.2860.

Hare, Richard M. 1970. Meaning and speech acts. The Philosophical Review 79, 3-24. doi: http://dx.doi.org/10.2307/2184066.

Hargreaves, David. 2005. Agency and intentional action in Kathmandu Newari. Himalayan Linguistics5: 1-48

Krifka, Manfred. 2008. Basic notions of information structure. Acta Linguistica Hungarica 55: 243-276.Krifka, Manfred. 2013. Response particles as propositional anaphors. In Snider, Todd Semantics and

Linguistic Theory (SALT) 23, 1-18. doi: http://dx.doi.org/10.3765/salt.v23i0.2676.Krifka, Manfred. 2014. Embedding illocutionary acts. In Roeper, Tom & Margaret Speas (eds.),

Recursion: Complexity in cognition. Springer, 125-155.http://amor.cms.hu-berlin.de/~h2816i3x/Publications/Krifka_EmbeddingIllocutionaryActs.pdf

Krifka, Manfred. 2015. Bias in Commitment Space Semantics: Declarative questions, negated questions, and question tag. SALT 25. LSA Open Journal Systems, 328-345.http://dx.doi.org/10.3765/salt

Krifka, Manfred. to appear. Negated polarity questions. In Lee, Chungmin & Ferenc Kiefer (eds.), Contrastiveness and Scalar Implicature. Berlin: Springer. http://amor.cms.hu-berlin.de/~h2816i3x/Publications/NegatedPolarityQuestions.pdf.

Ladd, Robert D. 1981. A first look at the semantics and pragmatics of negative questions and tag questions. Chicago Linguistic Society 17, 164-171. Chicago.

Lauer, Sven. 2013. Towards a dynamic pragmatics. Doctoral dissertation, Stanford University.Li, Charles & Sandra Thompson. 1981. Mandarin Chinese: A Functional Reference Grammar.

Berkeley: University of California Press.

References: References 38 / 39

Repp, Sophie. 2013. Common ground management: Modal particles, illocutionary negation and VERUM. In Gutzmann, Daniel & Hans-Martin Gärtner (eds.), Beyond Expressives – Explorations in Use-Conditional Meaning. 231-274. Leiden: Brill. doi: http://dx.doi.org/10.1163/9789004183988_008.

Romero, Maribel & Chung-hye Han. 2004. On negative yes/no questions. Linguistics and Philosophy 27, 609-658. doi: http://dx.doi.org/10.1023/b:ling.0000033850.15705.94.

Romero, Maribel. 2005. Two approaches to biased yes/no questions. WCCFL 24, 352-360. Searle, John R. & Daniel Vanderveken. 1985. Foundations of Illocutionary Logic. Cambridge:

Cambridge University Press.Speas, Margaret. 2004. Evidentiality, logophoricity and the syntactic representation of pragmatic

features. Lingua 114, 255-276. doi: http://dx.doi.org/10.1016/s0024-3841(03)00030-5.Szabolcsi, Anna. 1982. Model theoretic semantics of performatives. In: Kiefer, Ferenc (ed.),

Hungarian Linguistics, 515-535. Amsterdam: John Benjamins. doi: http://dx.doi.org/10.1075/llsee.4.16sza.

Truckenbrodt, Hubert. 2006. On the semantic motivation of syntactic verb movement to C in German.Theoretical Linguistics 32: 257-306. doi: http://dx.doi.org/10.1515/tl.2006.018.

Truckenbrodt, Hubert. 2015. Intonation phrases and speech acts. In: Kluck, Marlies, Dennis Ott & Mark de Vries (eds.), Parenthesis and ellipsis: Crosslinguistic and theoretical perspectives, 301-349. Berlin: Walter de Gruyter. doi: 4 9 2 285 8:3 + .

Wechsler, Stephen. 2015. Self-ascription in conjunct-disjunct systems. In: Floyd, Simeon, ElisabethNorcliffe & Lila San Roque, (eds), Egophoricity. Amsterdam: John Benjamins.

References: References 39 / 39